Primary brain tumors are the third leading contributor to cancer-related mortality in young adults, are the second leading contributor in children, and appear to be increasing in incidence both in the pediatric and geriatric population1-4. Gliomas are the most common type of primary brain tumors; 20,000 cases are diagnosed and 14,000 glioma-related deaths occur annually in the United States5-8. Gliomas are heterogeneous with respect to their malignant behavior and, in their most common and aggressive forms, anaplastic astrocytoma (AA-grade III) and glioblastoma multiforme (GBM-grade IV), are rapidly progressive and nearly uniformly lethal9; 10. Currently available therapeutic modalities have minimal curative potential for these high-grade tumors and often exacerbate the already severe morbidities imposed by their location in the central nervous system. Thus patients with malignant glioma are often struck in the most productive period of their lives; frequent deterioration of mental faculties and a high case:fatality ratio contribute to the unique personal and social impact of these tumors.
The cornerstones of oncologic management of malignant glioma are resection and radiation therapy11-16. With modern surgical and radiotherapeutic techniques the mean duration of survival has increased to 82 weeks for glioblastoma multiforme and 275 weeks for anaplastic astrocytoma, although 5-year survival rates have only increased from 3 to 6% for glioblastoma multiforme and 12.1% for anaplastic astrocytoma6-8. The major prognostic indicators for prolonged survival are younger age (<40 yrs) and performance status (KPS score>70)17. Resections of >90% of bulky tumors are usually attempted provided that vital functional anatomy is spared. When used in conjunction with post-operative radiation therapy, the impact of extent of resection on duration of survival is less clear18; 19. The addition of chemotherapy to resection and radiation provides only marginal survival advantage to patients with anaplastic astrocytoma or glioblastoma multiforme20-23. Nitrosureas alone or in combination with procarbazine and vincristine are the conventional drugs used in the community and appear to improve the 1-year and 2-year survival rates by 15% without impacting on the overall median survival24; 25. More aggressive regimens incorporating platinum-based drugs and topoisomerase inhibitors are under investigation26. The role of high-dose chemotherapy with stem cell rescue has not been substantiated to date27-29.
Approximately 80% of recurrent tumors arise from radiographically enhancing remnants of the original incompletely resected tumor10; 30; 31. Provided recurrences are unifocal and amenable in their location to aggressive re-resection, this approach can extend survival duration, particularly for patients with anaplastic astrocytoma and those glioblastoma multiforme patients with a KPS>70.10 The median survival of recurrent glioblastoma multiforme patients treated with re-resection is 36 weeks10; 30; 31. Radiation therapy in the form of either brachytherapy or stereotactic radiosurgery may extend the duration of survival in re-resected recurrent glioblastoma multiforme patients by only 10-12 weeks32. The use of chemotherapy in the setting of recurrent disease should be in the context of available clinical trials, as its efficacy in this patient population is unsubstantiated.
The continued dismal prognosis of malignant glioma has prompted the clinical investigation of novel therapeutic entities, including, but not limited to: gene therapy (TK-suicide, antisense inhibition of tumor growth factor receptors, conditionally lethal viral vectors), immunotherapy (antibody, tumor cell vaccines, immunotoxins, adoptive transfer of activated lymphocytes), and anti-angiogenesis approaches33-40. The multiplicity of challenges faced in the development of effective adjuvant therapies for malignant glioma include the extensive infiltrative growth of tumor cells into normal brain parenchyma, the capacity of soluble factors elaborated from these tumors to attenuate the development of immune responses, and the difficulty of establishing clinically meaningful therapeutic ratios when administering therapeutics into the central nervous system (CNS). Early clinical evaluation of novel therapeutics is clearly indicated in this patient population.
Recently, receptors for transferrin and growth factors have been the subject of experimental glioma therapeutics utilizing ligands for these receptors conjugated to toxins or radionucleotides as a delivery system41. The specificity of this approach relies on the unique expression or over-expression of targeted receptors on glioma cells compared to normal brain. Interestingly, some receptor complexes for interleukins utilized by the immune system are expressed by gliomas, in particular high-affinity IL-13 receptors42-48. Unlike the IL-13 receptor trimolecular complex utilized by the immune system, which consists of the IL-13Rα1, the IL-4Rβ, and γc, glioma cells overexpress a unique IL-13Rα2 chain capable of binding IL-13 independently of the requirement for IL-4Rβ or γc44; 49; 50. Like its homologue IL-4, IL-13 has pleotrophic immunoregulatory activity outside the CNS51-53. Both cytokines stimulate IgE production by B lymphocytes and suppress pro-inflammatory cytokine production by macrophages. The immunobiology of IL-13 within the CNS is largely unknown.
Detailed studies by Debinski et al. using autoradiography with radiolabeled IL-13 have demonstrated abundant IL-13 binding on nearly all malignant glioma tissues studied42; 45; 46; 48. Moreover, the binding is highly homogeneous within tumor sections and from single cell analysis46; 48. Scatchard analyses of IL-13 binding to human glioma cell lines reveals on average 17,000-28,000 binding sites/cell45. Molecular analysis using probes specific for IL-13Rα2 mRNA fail to demonstrate expression of the glioma-specific receptor by normal brain elements in all CNS anatomic locations42; 43. Furthermore, autoradiography with radiolabeled IL-13 failed to demonstrate detectable specific IL-13 binding in the CNS, suggesting that the shared IL13Rα1/IL-4β/γc receptor is also not expressed at detectable levels in the CNS46. These findings were independently verified using immunohistochemical techniques on non-pathologic brain sections with antibodies specific for IL-13Rα1 and IL-4β54. Thus IL-13Rα2 stands as the most specific and ubiquitously expressed cell-surface target for glioma described to date.
As a strategy to exploit the glioma-specific expression of IL-13Rα2 to the CNS, molecular constructs of the IL-13 cytokine have been described that fuse various cytotoxins (Pseudomonas exotoxin and Diptheria toxin) to its carboxyl terminal55-58. Internalization of these toxins upon binding to IL-13 receptors is the basis of the selective toxicity of these fusion proteins. These toxins display potent cytotoxicity towards glioma cells in vitro at picomolar concentrations55. Human intracranial glioma xenografts in immunodeficient mice can be eliminated by intratumor injection of the IL-13-toxin fusion protein without observed toxicities55. These studies support the initiation of clinical investigation utilizing IL-13-directed immunotoxins loco-regionally for malignant glioma.
However, the binding of IL-13-based cytotoxins to the broadly expressed IL-13Rα1/IL-4β/γc receptor complex has the potential of mediating untoward toxicities to normal tissues outside the CNS, and thus limits the systemic administration of these agents. IL-13 has been extensively dissected at the molecular level: structural domains of this cytokine that are important for associating with individual receptor subunits have been mapped55; 58. Consequently, selected amino acid substitutions in IL-13 have predictable effects on the association of this cytokine with its receptor subunits. Amino acid substitutions in IL-13's alpha helix A, in particular at amino acid 13, disrupt its ability to associate with IL-4β, thereby selectively reducing the affinity of IL-13 to the IL-13Rα1/IL-4β/γc receptor by a factor of five55; 57; 58. Surprisingly, binding of mutant IL-13(E13Y) to IL-13Rα2 was not only preserved but increased relative to wild-type IL-13 by 50-fold. Thus, minimally altered IL-13 analogs can simultaneously increase IL-13's specificity and affinity for glioma cells via selective binding to IL-13Rα2 relative to normal tissues bearing IL-13Rα1/IL-4β/γc receptors.
Malignant gliomas represent a clinical entity that is highly attractive for immunotherapeutic intervention since 1) most patients with resection and radiation therapy achieve a state of minimal disease burden and 2) the anatomic location of these tumors within the confines of the CNS make direct loco-regional administration of effector cells possible. At least two pathologic studies have demonstrated that the extent of perivascular lymphocytic infiltration in malignant gliomas correlates with an improved prognosis59-61. Animal model systems have established that glioma-specific T cells, but not lymphokine-activated killer (LAK) cells, can mediate the regression of intracerebrally implanted gliomas62-71. T cells, unlike LAK cells, have the capacity to infiltrate into brain parenchyma and thus can target infiltrating tumor cells that may be distant from the primary tumor. Despite these findings, there is a substantial body of evidence that gliomas actively subvert immune destruction, primarily by the elaboration of immunosuppressive cytokines (TGF-β2) and prostaglandins, which, inhibit the induction/amplification of glioma-reactive T cell responses72-74. These findings have prompted the evaluation of ex vivo expanded anti-glioma effector cells for adoptive therapy as a strategy to overcome tumor-mediated limitations of generating responses in vivo.
At least ten pilot studies involving the administration of ex vivo activated lymphocytes to malignant glioma resection cavities have been reported to date75-85. Despite the variety of effector cell types (LAK, TILs, alloreactive CTLs), their heterogeneous composition/variability of composition from patient to patient, and the often modest in vitro reactivity of these effector cells towards glioma targets, these studies, in aggregate, report an approximate 50% response rate in patients with recurrent/refractory disease with anecdotal long-term survivors. These studies support the premise that a superior clinical effect of cellular immunotherapy for glioma might be expected with homogenous highly potent effector cells.
These pilot studies also report on the safety and tolerability of direct administration of ex vivo activated lymphocytes and interleukin-2 (IL-2), a T cell growth factor, into the resection cavity of patients with malignant glioma75; 76; 78; 82; 86-92. Even at large individual cell doses (>109 cells/dose), as well as high cumulative cell doses (>27×109 cells), toxicities are modest, and typically consist of grade II or less transient headache, nausea, vomiting and fever. As noted above, these studies also employed the co-administration of rhIL-2 to support the in vivo survival of transferred lymphocytes. Multiple doses given either concurrently with lymphocytes or sequentially after lymphocyte administration were tolerated at doses as high as 1.2×106 IU/dose for 12-dose courses of IL-2 delivered every 48-hours.
Based on the findings outlined above, strategies to improve the anti-tumor potency of lymphocyte effector cells used in glioma immunotherapy are under development. One approach utilizes bi-specific antibodies capable of co-localizing and activating T lymphocytes via an anti-CD3 domain with glioma targets utilizing an epidermal growth factor receptor (EGFR) binding domain93-96. Preliminary clinical experience with this bi-specific antibody in combination with autologous lymphocytes suggests that T cells are activated in situ in the resection cavity. Targeting infiltrating tumor cells within the brain parenchyma, however, is a potentially significant limitation of this approach. T cells might have significantly increased anti-glioma activity if they are specific for target antigens expressed by gliomas. A growing number of human genes encoding tumor antigens to which T lymphocytes are reactive have been cloned, including the SART-1 gene, which appears to be expressed by nearly 75% of high-grade gliomas97. Both dendritic cell-based in vitro cell culture techniques, as well as tetramer-based T cell selection technologies are making feasible the isolation of antigen-specific T cells for adoptive therapy. Since antigens like SART-1 are recognized by T cells in the context of restricting HLA alleles, antigen-specific approaches will require substantial expansion in the number of antigens and restricting HLA alleles capable of presenting these antigens to be broadly applicable to the general population of glioma patients.
Chimeric antigen receptors engineered to consist of an extracellular single chain antibody (scFvFc) fused to the intracellular signaling domain of the T cell antigen receptor complex zeta chain (scFvFc:ζ) have the ability, when expressed in T cells, to redirect antigen recognition based on the monoclonal antibody's specificity98. The design of scFvFc:ζ receptors with target specificities for tumor cell-surface epitopes is a conceptually attractive strategy to generate antitumor immune effector cells for adoptive therapy as it does not rely on pre-existing anti-tumor immunity. These receptors are “universal” in that they bind antigen in a MHC independent fashion, thus, one receptor construct can be used to treat a population of patients with antigen-positive tumors. Several constructs for targeting human tumors have been described in the literature including receptors with specificities for Her2/Neu, CEA, ERRB-2, CD44v6, and epitopes selectively expressed on renal cell carcinorna98-104. These epitopes all share the common characteristic of being cell-surface moieties accessible to scFv binding by the chimeric T cell receptor. In vitro studies have demonstrated that both CD4+ and CD8+ T cell effector functions can be triggered via these receptors. Moreover, animal models have demonstrated the capacity of adoptively transferred scFvFc:ζ expressing T cells to eradicate established tumors105. The function of primary human T cells expressing tumor-specific scFvFc:ζ receptors have been evaluated in vitro; these cells specifically lyse tumor targets and secrete an array of pro-inflammatory cytokines including IL-2, TNF, IFN-γ, and GM-CSF104. Phase I pilot adoptive therapy studies are underway utilizing autologous scFvFc:ζ-expressing T cells specific for HIV gp120 in HIV infected individuals and autologous scFvFc:ζ-expressing T cells with specificity for TAG-72 expressed on a variety of adenocarcinomas, including breast and colorectal adenocarcinoma.
Investigators at City of Hope have engineered a CD20-specific scFvFc:ζ receptor construct for the purpose of targeting CD20+ B-cell malignancy and an L1-CAM-specific chimeric immunoreceptor for targeting neuroblastoma106. Preclinical laboratory studies have demonstrated the feasibility of isolating and expanding from healthy individuals and lymphoma patients CD8+ CTL clones that contain a single copy of unrearranged chromosomally integrated vector DNA and express the CD20-specific scFvFc:ζ receptor107. To accomplish this, purified linear plasmid DNA containing the chimeric receptor sequence under the transcriptional control of the CMV immediate/early promoter and the NeoR gene under the transcriptional control of the SV40 early promoter was introduced into activated human peripheral blood mononuclear cells by exposure of cells and DNA to a brief electrical current, a procedure called electroporation. Utilizing selection, cloning, and expansion methods currently employed in FDA-approved clinical trials at the Fred Hutchinson Cancer Research Center, Seattle, Wash., gene modified CD8+ CTL clones with CD20-specific cytolytic activity have been generated from each of six healthy volunteers in 15 separate electroporation procedures. These clones when co-cultured with a panel of human CD20+ lymphoma cell lines proliferate, specifically lyse target cells, and are stimulated to produce cytokines: